From c93131cef6dbb4e415fd2b3625f644c6714e7465 Mon Sep 17 00:00:00 2001
From: Ed Addario <eaddario@hotmail.com>
Date: Fri, 10 Oct 2025 13:26:51 +0100
Subject: [PATCH] Remove --no-bias option

---
 include/llama.h             | 1 -
 src/llama-quant.cpp         | 3 +--
 tools/quantize/quantize.cpp | 6 +-----
 3 files changed, 2 insertions(+), 8 deletions(-)

diff --git a/include/llama.h b/include/llama.h
index 16f6124727..1df8f96920 100644
--- a/include/llama.h
+++ b/include/llama.h
@@ -365,7 +365,6 @@ extern "C" {
         void * tensor_types;                  // pointer to vector containing tensor types
         void * prune_layers;                  // pointer to vector containing layer indices to prune
         float target_bpw;                     // target bits per weight (bpw)
-        bool no_bias;                         // use mean square error estimation only (no aligment bias)
     } llama_model_quantize_params;
 
     typedef struct llama_logit_bias {
diff --git a/src/llama-quant.cpp b/src/llama-quant.cpp
index 7b3e956193..4ad5124d1a 100644
--- a/src/llama-quant.cpp
+++ b/src/llama-quant.cpp
@@ -2180,8 +2180,7 @@ llama_model_quantize_params llama_model_quantize_default_params() {
         /*.kv_overrides                =*/ nullptr,
         /*.tensor_type                 =*/ nullptr,
         /*.prune_layers                =*/ nullptr,
-        /*.target_bpw                  =*/ -1.0f,
-        /*.no_bias                     =*/ false
+        /*.target_bpw                  =*/ -1.0f
     };
 
     return result;
diff --git a/tools/quantize/quantize.cpp b/tools/quantize/quantize.cpp
index d355f97274..c254c3f6b2 100644
--- a/tools/quantize/quantize.cpp
+++ b/tools/quantize/quantize.cpp
@@ -118,7 +118,7 @@ static bool try_parse_ftype(const std::string & ftype_str_in, llama_ftype & ftyp
 [[noreturn]]
 static void usage(const char * executable) {
     printf("usage: %s [--help] [--allow-requantize] [--leave-output-tensor] [--pure] [--imatrix] [--include-weights] [--exclude-weights]\n", executable);
-    printf("       [--target-bpw n] [--no-bias] [--output-tensor-type] [--token-embedding-type] [--tensor-type] [--prune-layers] [--keep-split] [--override-kv]\n");
+    printf("       [--target-bpw n] [--output-tensor-type] [--token-embedding-type] [--tensor-type] [--prune-layers] [--keep-split] [--override-kv]\n");
     printf("       model-f32.gguf [model-quant.gguf] type [nthreads]\n\n");
     printf("  --allow-requantize: allows requantizing tensors that have already been quantized. Warning: This can severely reduce quality compared to quantizing from 16bit or 32bit\n");
     printf("  --leave-output-tensor: will leave output.weight un(re)quantized. Increases model size but may also increase quality, especially when requantizing\n");
@@ -134,8 +134,6 @@ static void usage(const char * executable) {
     printf("      Advanced option to remove all tensors from the given layers\n");
     printf("  --target-bpw: target bits per weight (bpw). Must be a positive number between 0.0 and 8.0\n");
     printf("      Advanced option to automatically select quantization types to achieve a total bits per weight (bpw) target\n");
-    printf("  --no-bias: use mean square error estimation only (no aligment bias)\n");
-    printf("      Advanced option use MSE only and disable aligment bias error estimation\n");
     printf("  --keep-split: will generate quantized model in the same shards as input\n");
     printf("  --override-kv KEY=TYPE:VALUE\n");
     printf("      Advanced option to override model metadata by key in the quantized model. May be specified multiple times.\n");
@@ -559,8 +557,6 @@ int main(int argc, char ** argv) {
             if (arg_idx == argc-1 || !parse_target_bpw(argv[++arg_idx], target_bpw)) {
                 usage(argv[0]);
             }
-        } else if (strcmp(argv[arg_idx], "--no-bias") == 0) {
-            params.no_bias = true;
         } else if (strcmp(argv[arg_idx], "--prune-layers") == 0) {
             if (arg_idx == argc-1 || !parse_layer_prune(argv[++arg_idx], prune_layers)) {
                 usage(argv[0]);